Hyundai Motor Group delays in-house autonomous driving to 2029, partners with Nvidia for 2028 launch

Hyundai Motor Group postpones proprietary autonomous driving launch

Hyundai Motor Group announced it will launch vehicles equipped with its proprietary driver-assistance software in late 2029, pushing back the timeline by two years from the original late-2027 target1

1

. The delay underscores significant challenges the South Korean automaker faces in developing automated driving software in-house while competing against Tesla and Chinese rivals in the race to commercialize vehicle autonomy2

2

. To bridge this gap, Hyundai Motor Group is partnering with Nvidia to accelerate deployment of advanced driver-assistance systems starting in 2028.

Nvidia partnership brings Level 2+ and Level 2++ systems by 2028

The automaker will roll out Level 2+ and Level 2++ systems powered by Nvidia's Hyperion 10 platform beginning in 20283

3

. Level 2+ enables hands-free driving on highways, while Level 2++ extends hands-free urban driving capabilities similar to Tesla's Full Self-Driving system, though both still require driver supervision1

1

. Park Min-woo, president at Hyundai Motor Group and former Nvidia executive who joined in January, emphasized that "our partnership with Nvidia is not about leaving our destiny entirely in their hands"2

2

. The vehicles will initially bypass expensive lidar sensors in favor of cameras, ultrasonic sensors and radar, though Hyundai is considering lidar sensors for Level 3 systems that allow hands-off driving under specific conditions1

1

.

Strategic shift reflects leadership changes and development challenges

Park's strategy marks a significant departure from his predecessor Song Chang-hyeon, who focused heavily on internal software development before his abrupt departure in December1

1

. The new two-track approach uses Nvidia's technology to accelerate commercialization while developing the proprietary Atria AI system to secure long-term technological independence3

3

. Park explained the 2029 timeline is "a strategic choice to focus on our North Star while using Nvidia's technology for mass production starting in 2028," deliberately avoiding rushing an immature system3

3

. Hyundai will co-design the technology with Nvidia and use data from the system to train and refine its own software platform, dubbed Atria2

2

.

Data Flywheel strategy leverages 7 million annual vehicle sales

Hyundai and affiliate Kia Corp, which together rank as the world's third-largest automaker with annual sales exceeding 7 million vehicles across about 190 countries, plan to leverage their global fleet scale through a Data Flywheel approach1

1

3

3

. This creates a continuous cycle of data collection, AI training, validation and deployment to harvest real-world driving data for training autonomous driving technology3

3

. The group currently operates about 40 dedicated data-collection vehicles and is standardizing sensor systems across Hyundai Motor, Kia, 42dot and Motional around Nvidia's DRIVE Hyperion 10 architecture3

3

. Hyundai expects to surpass competitors in accumulated driving data by 2033, with one executive suggesting they could close the gap within five years once mass production begins1

1

3

3

.

Expanding Nvidia collaboration beyond autonomous vehicles

Hyundai Motor Group is deepening collaboration with Nvidia spanning autonomous driving, AI data centers and humanoid robots developed by Hyundai-owned Boston Dynamics1

1

. Park emphasized that "competitiveness is determined by how much data you secure, how quickly you learn and how effectively you can reflect those results in actual products and services"3

3

. The automaker's 42dot subsidiary is also developing vision-language-action models alongside its end-to-end technology, with plans to begin real-vehicle testing late this year through early next year3

3

. Additionally, Hyundai plans to conduct a real-world Level 4 autonomous driving pilot in Gwangju by the end of this year in partnership with the Ministry of Land, Infrastructure and Transport3

3

. Park noted that "to truly ensure safety, we believe sensor redundancy is critical: if one sensor fails, others must be able to maintain safe operation," though no timeline was provided for Level 3 systems commercialization1

1

.